Register a SA Forums Account here!
JOINING THE SA FORUMS WILL REMOVE THIS BIG AD, THE ANNOYING UNDERLINED ADS, AND STUPID INTERSTITIAL ADS!!!

You can: log in, read the tech support FAQ, or request your lost password. This dumb message (and those ads) will appear on every screen until you register! Get rid of this crap by registering your own SA Forums Account and joining roughly 150,000 Goons, for the one-time price of $9.95! We charge money because it costs us money per month for bills, and since we don't believe in showing ads to our users, we try to make the money back through forum registrations.
 
  • Post
  • Reply
TomR
Apr 1, 2003
I both own and operate a pirate ship.

Putting the TIM on the CPU before it's in the socket is like standing up to wipe.

Adbot
ADBOT LOVES YOU

Paul MaudDib
May 2, 2006

"Tell me of your home world, Usul"


Paul MaudDib fucked around with this message at 18:05 on Sep 19, 2017

BangersInMyKnickers
Nov 3, 2004

I have a thing for courageous dongles



I have seen at least 3 different places do a similar test and came to the conclusion that the X is the best application pattern

e: Apply the TIM while out of the socket also has a risk of smearing paste on the underside if you have messy fingers.

BangersInMyKnickers fucked around with this message at 18:15 on Sep 19, 2017

SoftNum
Mar 31, 2011



Also not all processors are drop and clamp installations. Notably:

https://www.youtube.com/watch?v=Zyik-vSfvwo

nerox
May 20, 2001



No idea how many times I have seen this and I just now noticed that syringe says Hellman's on it.



My father in law makes his drinks by doing liquor, then mixer, then puts ice in the cup and splashes the drink on the counter.

CaptainSarcastic
Jul 6, 2013

HAIL SATAN




nerox posted:

No idea how many times I have seen this and I just now noticed that syringe says Hellman's on it.

I had not caught that before - I love that image.

PerrineClostermann
Dec 15, 2012

by FactsAreUseless


Is there any risk of some sort of weird air pocket/bubble existing betwen the CPU and cooler if you manually spread the paste beforehand?

Rexxed
May 1, 2010

Dis is amazing!
I gotta try dis!



PerrineClostermann posted:

Is there any risk of some sort of weird air pocket/bubble existing betwen the CPU and cooler if you manually spread the paste beforehand?

Yes, that's why you don't do that.

BangersInMyKnickers
Nov 3, 2004

I have a thing for courageous dongles



Air bubbles could still be in the compound but not exposed, as demonstrated in that link where they tested it on the plexiglass plate. The X pattern gives a place for the air to vent out as the TIM pushes in and thins.

PerrineClostermann
Dec 15, 2012

by FactsAreUseless


Rexxed posted:

Yes, that's why you don't do that.

That's what I thought, but couldn't remember if I'd imagined it or not.

SoftNum
Mar 31, 2011



the proper technique:

https://www.youtube.com/watch?v=kSCVahyh3Dk

CaptainSarcastic
Jul 6, 2013

HAIL SATAN



I, too, always preferred using Colgate for my thermal paste needs.

EmpyreanFlux
Mar 1, 2013

The AUDACITY! The IMPUDENCE! The unabated NERVE!



I bet MSI had no idea this would become an instant classic

repiv
Aug 13, 2009



Looks like Intels promise to open up Thunderbolt has come to fruition, Gigabyte just announced an X399 board with support for TB3 cards: https://www.anandtech.com/show/1184...99-designare-ex

wargames
Mar 16, 2008

official yospos cat censor


repiv posted:

Looks like Intels promise to open up Thunderbolt has come to fruition, Gigabyte just announced an X399 board with support for TB3 cards: https://www.anandtech.com/show/1184...99-designare-ex

royalty free is awesome!

SwissArmyDruid
Feb 14, 2014



Paul MaudDib posted:

If Epyc has both stereoisomers wouldn't they need to have both, otherwise two of the dies would be broken?

Plausible though, I hadn't thought about that

Back to the "glass" thing, it seems a marketing drone playing a one-person game of Telephone in their head off something an engineer said seems to be the culprit.

The two extra dies are dead. As in dead-dead. There will be no Athlon/Phenom core unlocking going on here.



https://twitter.com/cavemanjim/status/909429198732668930

SwissArmyDruid fucked around with this message at 02:08 on Sep 20, 2017

GRINDCORE MEGGIDO
Feb 28, 1985




I'm totally surprised they had enough dead dies to make that worthwhile.

SwissArmyDruid
Feb 14, 2014



Not sure if srs...

GRINDCORE MEGGIDO
Feb 28, 1985




SwissArmyDruid posted:

Not sure if srs...

srs

I suppose tho I could totally be wrong and they're churning out tons of chips and relatively few TR packages.

On the substrate, I wonder if they're even connected at all to the PCB or just pads on there.
Cos if the wiring is hooked up it'd be another vector for package faults.

EmpyreanFlux
Mar 1, 2013

The AUDACITY! The IMPUDENCE! The unabated NERVE!


Not hooked up, and the problem with that twitter guys diagram is those destined for the landfill are instead used as spacers for Threadripper. Like, that's where all of AMDs dead dies go, they're probably close to 100% usage of everything GloFo gives them.

underage at the vape shop
May 11, 2011

by Cyrano4747


post so i can find a post

Watermelon Daiquiri
Jul 10, 2010




GRINDCORE MEGGIDO posted:

srs

I suppose tho I could totally be wrong and they're churning out tons of chips and relatively few TR packages.

On the substrate, I wonder if they're even connected at all to the PCB or just pads on there.
Cos if the wiring is hooked up it'd be another vector for package faults.

14nm success rate for something that's not on a cheaper, fewer mask, process is ~75%. That includes ALL dies which pass, whether its bin 1 or one that fails one thing like voltage, but not so egregiously that it would take 150W to hit 3.8ghz or whatever. For zen, the dies are 213mm2, which means there are approx. 300 dies on a wafer: 225 that pass, 75 that outright fail. I think there's more than enough dead dies for threadripper Hell, even if they got things going amazingly up to an 85% yield (the highest i've personally seen for a 14nm product was 86-88% for a 'budget' arm soc), there'd still be enough for like 20 tr processors per wafer

Watermelon Daiquiri fucked around with this message at 03:21 on Sep 20, 2017

GRINDCORE MEGGIDO
Feb 28, 1985




Watermelon Daiquiri posted:

14nm success rate for something that's not on a cheaper, fewer mask, process is ~75%. That includes ALL dies which pass, whether its bin 1 or one that fails one thing like voltage but not so egregiously that it would take 150W to hit 3.8ghz or whatever. For zen, the dies are 213mm/super2[/super], which means theres approx. 300 dies on a wafer: 225 that pass, 75 that outright fail. I think there's more than enough dead dies for threadripper Hell, even if they got things going amazingly up to an 85% yield (the highest i've personally seen for a 14nm product was 86-88% for a 'budget' arm soc), there'd still be enough for like 20 tr processors per wafer

Fair enough. This makes sense, thank you.

Watermelon Daiquiri
Jul 10, 2010




Looking up the die size allowed me to come across this amazing site which goes in depth into the zen architecture as well as others, apparently: https://en.wikichip.org/wiki/amd/microarchitectures/zen

underage at the vape shop
May 11, 2011

by Cyrano4747


whats the least problematic ryzen ram? What guides are there to get it to work?

I'm getting sick of it not posting 5 times, then having to go into the bios, change nothing, restart, just to get it to run 2933 on 3200 QVL approved ram. THis is every single time I turn my PC on.

wargames
Mar 16, 2008

official yospos cat censor


Watermelon Daiquiri posted:

Looking up the die size allowed me to come across this amazing site which goes in depth into the zen architecture as well as others, apparently: https://en.wikichip.org/wiki/amd/microarchitectures/zen

Can this be put in the op, like really this needs be put in the op so its never lost.

Anime Schoolgirl
Nov 28, 2002






underage at the vape shop posted:

whats the least problematic ryzen ram? What guides are there to get it to work?

I'm getting sick of it not posting 5 times, then having to go into the bios, change nothing, restart, just to get it to run 2933 on 3200 QVL approved ram. THis is every single time I turn my PC on.
You might have to set the clocks/voltage manually and loosen timings, it might just be trying to do the XMP profile each bootup despite it obviously not going to work with Ryzen in most cases. I've had a similar problem with an Asus B350 Prime, but with non B-die ram.

There isn't a comprehensive list of ram that works with Ryzen because there's just thousands of SKUs, but there's some known Samsung B-die ram sticks. Unsurprisingly their prices are inflated.
https://www.reddit.com/r/Amd/commen...onfusion_eg_on/

New Zealand can eat me
Aug 29, 2008



TomR posted:

Putting the TIM on the CPU before it's in the socket is like standing up to wipe.

underage at the vape shop
May 11, 2011

by Cyrano4747


Anime Schoolgirl posted:

You might have to set the clocks/voltage manually and loosen timings, it might just be trying to do the XMP profile each bootup despite it obviously not going to work with Ryzen in most cases. I've had a similar problem with an Asus B350 Prime, but with non B-die ram.

There isn't a comprehensive list of ram that works with Ryzen because there's just thousands of SKUs, but there's some known Samsung B-die ram sticks. Unsurprisingly their prices are inflated.
https://www.reddit.com/r/Amd/commen...onfusion_eg_on/

yeah i spent all day loving around with timings and said fuckit and got the expensive rear end gskill made for ryzen ram and with XMP I got 3200MHZ first boot

its weird having a functional computer

Watermelon Daiquiri
Jul 10, 2010





Going by this logic, putting goop on the cpu while its in the socket is wiping back to front

TomR
Apr 1, 2003
I both own and operate a pirate ship.

Watermelon Daiquiri posted:

Going by this logic, putting goop on the cpu while its in the socket is wiping back to front

The MSI video is wiping back to front.

New Zealand can eat me
Aug 29, 2008



Apparently 14+ is turning into 12nm for both vega and ryzen? http://www.tomshardware.com/news/am...2018,35502.html

Seamonster
Apr 30, 2007

IMMER SIEGREICH


So uh... 12nm node is coming.

EmpyreanFlux
Mar 1, 2013

The AUDACITY! The IMPUDENCE! The unabated NERVE!


New Zealand can eat me posted:

Apparently 14+ is turning into 12nm for both vega and ryzen? http://www.tomshardware.com/news/am...2018,35502.html

I wonder if they could get something like 50% of the benefits of moving to "10nm" and this is why Nvidia and AMD basically laughed off 10nm for their products. "Hmm, yes redesign everything, new masks included, for a nearly dead node or, or, now bear with me, we just mature the current node further and spend a gigantic assload less for making it halfway."

Seamonster
Apr 30, 2007

IMMER SIEGREICH


I am salivating for Ryzen with a few percent IPC improvement, baked in errata fixes and OC to 4.4-4.5 Ghz. Oh and less finicky memory support.

This is the right call to get product out to market faster.

The dumpster fire known as Vega on the other hand may get the flames put out with this new node? (It will still be a disgusting smelly mess, just no longer actively burning?)

EdEddnEddy
Apr 5, 2012





So I am still catching up on the thread (page 38 still) but wanted to check. Anyone see that Tesla appears to be working with AMD to make an AI chip for their cars?

The stock is finally back over $14 and rising.

eames
May 9, 2009



EV owners rejoice, no more cold cabins in winter

ufarn
May 30, 2009


Can't wait to buy three Teslas and just have them sit in the garage and mine bitcoins.

Malcolm XML
Aug 8, 2009

I always knew it would end like this.


EdEddnEddy posted:

So I am still catching up on the thread (page 38 still) but wanted to check. Anyone see that Tesla appears to be working with AMD to make an AI chip for their cars?

The stock is finally back over $14 and rising.

hell yeah bois up uP UP!!!!


FaustianQ posted:

I wonder if they could get something like 50% of the benefits of moving to "10nm" and this is why Nvidia and AMD basically laughed off 10nm for their products. "Hmm, yes redesign everything, new masks included, for a nearly dead node or, or, now bear with me, we just mature the current node further and spend a gigantic assload less for making it halfway."

nodes are marketing names at this point this is very likely "10nm" FEOL with "14nm" BEOL or somesuch like tsmc 16nm used older BEOL tech

Heck intel is basically hosed with 10nm, 14+++ is gonna be better than 10 for everything but extreme density/power needs (apparently SemiAccurate has the scoop on whats wrong but i am not spending 1k on that)

Adbot
ADBOT LOVES YOU

SwissArmyDruid
Feb 14, 2014



Malcolm XML posted:

hell yeah bois up uP UP!!!!


nodes are marketing names at this point this is very likely "10nm" FEOL with "14nm" BEOL or somesuch like tsmc 16nm used older BEOL tech

Heck intel is basically hosed with 10nm, 14+++ is gonna be better than 10 for everything but extreme density/power needs (apparently SemiAccurate has the scoop on whats wrong but i am not spending 1k on that)

Jesus H. Chrst on a pogo stick with Mary and Judas, a thousand bucks for paywall? I mean, I think Semi-accurate is legit and all, but what the hell?

  • 1
  • 2
  • 3
  • 4
  • 5
  • Post
  • Reply